Think about the last post you were genuinely proud of.

The research ate a weekend. The draft took three evenings. You rewrote the intro four times, fixed the screenshots, added a comparison table, and hit publish.

For most of the last twenty years, that effort came with a quiet promise. Google and other search engines would notice your work, readers would arrive, a few would subscribe, a few would click an ad or an affiliate link, and the next post would feel a little easier to write.

That deal, where you give the web something useful and the web sends people your way, built the independent internet. In 2026, it is coming apart. And the people who held up their end of it are the ones paying for it.

The bargain that built the web.

Blogging never made sense as pure charity. It worked because the arrangement was fair enough. Search engines crawled your pages, summarized them in a snippet, and sent readers to the full article. You got attention. They got a better product.

Everyone’s incentives was lined up.

Cloudflare, which sits in front of roughly a fifth of the web, has put numbers on how lopsided that arrangement has become. Its crawl-to-referral ratio measures how many pages a company’s bots fetch from websites for every visitor that company sends back.

In mid-2025, Cloudflare reported Google at roughly 14 crawls per referral, OpenAI at about 1,700, and Anthropic at about 73,000. Those figures have moved since. Cloudflare Radar data from early 2026 put Anthropic’s ClaudeBot near 24,000 to one in the first quarter and about 11,000 to one by early summer, while OpenAI sat around 857 to one and Google near five. The trend is improving. But a gap measured in thousands is not a partnership. It is a one-way pipe.

To be fair to the technology, some of those bots exist purely to train models and were never built to send anyone anywhere. That is exactly the point. The old handshake had a second half, and a lot of the new machinery simply does not have one.

The click is disappearing.

The clearest evidence of how reader behavior has changed comes from Pew Research Center. It tracked the actual browsing of about 900 U.S. adults in March 2025, covering roughly 68,000 Google searches. About 58% of those people ran at least one search that produced an AI-generated summary.

When no AI summary appeared, users clicked a regular result link on about 15% of visits. When one did appear, that dropped to about 8%.

Nearly half the clicks, gone.

Then there is the number that should bother every writer: only about 1% of visits to a page with an AI summary included a click on a source cited inside that summary. The citations are there. Readers just are not following them.

The summaries also seem to be doing their job of ending the search.

Pew found that 26% of visits with an AI summary ended the browsing session entirely, compared with 16% for ordinary results.

The reader got an answer and left. They never met the person who wrote it.

Google disputes the study’s methodology and says AI features help people discover a wider range of sites. It also noted that more than a billion people had used AI Overviews. Both things can be true: people may genuinely like the experience, and the writers who made it possible may still be losing out.

What the traffic charts look like now?

The picture from publishers’ own analytics is just as sobering.

Chartbeat, an analytics firm that tracks traffic across thousands of publisher sites, measured Google Search referrals to news and media sites falling about 33% worldwide between November 2024 and November 2025. The Reuters Institute surveyed 280 media leaders across 51 countries for its 2026 trends report, and they expected search traffic to fall by another 43% on average over the next three years.

The latest data is worse. Chartbeat’s 2026 Publisher Playbook reports that Google Search traffic to publishers dropped 40.2% year on year between July 2025 and July 2026. The year before, the drop was 21.9%. In other words, the decline nearly doubled. Google Discover, long the favorite backup channel for a lot of writers, saw its own decline widen from 6.6% to 34.3%. Search’s share of all pageviews on the network slid from about 9% in July 2024 to about 5% in July 2026.

Now the part that matters most if you run a small site. Chartbeat data reported by Axios in March 2026 showed that small publishers lost around 60% of their search traffic over two years, a sharper fall than any other category. Big brands with loyal audiences are weathering this far better than independents.

There is one more detail worth knowing. The Reuters Institute noted that publishers in lifestyle and utility categories, think weather, TV guides, horoscopes and how-to content, appeared to be hit hardest. That is the kind of plain informational writing that fills most blogs. It is also exactly what an AI summary can compress into three sentences.

And the replacement traffic is not coming. According to the same reporting, ChatGPT accounted for around 0.02% of publisher referrals, against about 7.3% for Google Search. Chatbots are answering the questions. They are not handing the audience back.

One honest caveat: nobody can say precisely how much of the decline is caused by AI. It is because search algorithm changes, social platform shifts and ordinary competition all play a part, and even Chartbeat has noted that the story is more nuanced than “AI killed traffic.” But the direction is not in doubt, and the timing lines up uncomfortably well with the arrival of AI summaries in 2024.

The money question.

Traffic is not just vanity. For most bloggers, traffic is rent, groceries and hosting. Display ads pay per impression. Affiliate links pay per click. Email lists grow from visitors. Remove the visitor and every one of those income streams weakens at once.

Meanwhile, the costs did not shrink. Hosting still renews. Tools still bill monthly. The hours it takes to research and write something properly have not dropped just because a machine can produce a passable imitation in seconds.

There is also an ugly side effect for site owners.

Cloudflare reported that automated traffic has now overtaken human traffic on its network, at around 57.5% of requests by mid-2026, and that AI “user action” crawling grew roughly fifteen-fold during 2025.

Bots do not click ads. They do cost bandwidth. Some publishers are now paying to serve their own work to machines that will never send a reader back.

Where the law is, and where it isn’t?

If you are waiting for the courts to settle this, brace for a long wait.

The biggest payout so far is Anthropic’s $1.5 billion settlement with authors, approved by a federal judge, which works out to roughly $3,000 per book across about 482,000 books, with more than 91% claimed. It was described as the largest copyright recovery in U.S. history. But look closely at what it covered. It dealt with books that were obtained from pirate libraries. The same litigation produced a ruling that training a model on lawfully acquired books can count as fair use. It was a settlement, so it sets no binding precedent, and it says nothing direct about blog posts scraped from the open web.

In other words, a blogger today has no equivalent claim and no clear legal route. Dozens of other AI copyright cases are still working through the courts, and nobody knows yet how they will land. So the realistic answer is that legal relief, if it comes, will come slowly. Bloggers need strategies that work this year.

How to fight back?

None of what follows is magic. Some of it will feel like extra work on top of work you already cannot keep up with. But together these moves shift you from depending on strangers’ algorithms toward owning your relationship with readers.

1. Build an audience you own.

This is the single most important move. An email list belongs to you. No summary box can sit between you and a subscriber. Offer something worth signing up for, whether that is a weekly note, a template, a checklist or a short course, and put the signup form where readers will actually see it. Even a list of a few hundred engaged people can outperform thousands of anonymous search visitors.

It is worth noting that Chartbeat’s own analysis found loyal, returning readers held steady even as search traffic fell. People who already know and like you are the audience AI summaries have the hardest time taking away.

2. Write what a summary cannot.

An AI can restate a definition.

It cannot tell the story of the time your server crashed at 2 a.m. during a client launch.

It cannot share the spreadsheet of your actual earnings, the photos from your actual test, or an opinion that costs you something to hold. Lean into first-hand experience, original data, interviews, case studies and a recognizable voice. Generic explainers are the easiest content to replace. A distinctive point of view is the hardest.

3. Go where the answer is not enough.

Some queries need depth, comparison, tools or community. Calculators, downloadable resources, templates, interactive guides and long-form tutorials give people a reason to visit that a three-sentence answer cannot satisfy. Ask yourself of every post: if someone got the short version, what would still make them click?

4. Diversify how people find you.

Putting everything on one search engine is the same mistake publishers made with social media a decade ago. Spread your presence across a newsletter, a YouTube channel or short video, a podcast, and one community where your readers already gather. Reuse a single piece of research across formats.

The Reuters Institute found that publishers themselves are moving effort away from traditional search optimization and toward direct relationships and video.

5. Diversify how you earn.

If ads and affiliate clicks are your only income, a traffic drop is a pay cut. Consider paid memberships, digital products, sponsored newsletter slots, consulting, workshops or freelance writing. Your blog can become the shop window rather than the entire shop.

6. Control who gets to crawl you.

You have more control than you might think. Your robots.txt file lets you disallow specific AI training crawlers, and many AI companies publish the names of their bots. Cloudflare now offers default blocking of AI crawlers for new customers, along with experiments in charging bots for access. Publishers are already returning more than a billion “payment required” responses per day to AI crawlers.

A warning, though. Google uses one crawler for both search indexing and its AI features, so blocking it outright can make you vanish from search. Google offers a separate opt-out token for training its Gemini models that does not remove you from Search, but it does not necessarily stop your content appearing in AI Overviews. Check each company’s current documentation before changing anything, and treat blocking as a tool, not a cure.

7. Still make yourself citable.

It can feel contradictory to block some bots while courting others, but reputation increasingly travels through citations. Clear headings, direct answers near the top, structured data, named authors with real bios and original statistics make your work easier to attribute. Even if only 1% of readers click a cited source, being the cited source builds brand recognition, and brand is what people later search for by name.

8. Support licensing and collective bargaining.

Individual bloggers have almost no leverage. Groups do. A licensing standard called Really Simple Licensing, backed by platforms including Reddit, Yahoo, Medium and Quora, aims to let sites set terms and fees for AI use of their content. It depends on AI companies choosing to cooperate, so it is early and uncertain.

But joining writer associations, supporting industry groups and making noise about attribution all help shift the default from “take freely” to “ask first.”

9. Measure what matters now.

Total pageviews are becoming a less honest scoreboard. Watch branded searches (people typing your name), direct visits, newsletter growth, returning-reader rate, and revenue per thousand visitors. A smaller audience that trusts you can be more valuable than a larger one that never remembers where it read the answer.

What this means for the next few years?

It would be dishonest to pretend every blogger will survive this. Thin sites, rewritten news and purely informational posts that exist to catch search traffic are in real danger.

Some will close, and more will quietly stop updating.

But the same pressure is clarifying what a blog is for. If a machine can summarize a topic, then the value of a human writer shifts toward trust, taste, experience and community. These are things you cannot scrape.

Several of the trends above, including the steady loyalty of returning readers, suggest that audiences still seek out individuals they believe in.

What bloggers are owed, and what they have mostly not received, is credit and compensation for the knowledge they supplied in the first place.

That fight is being waged in courts, in standards bodies and in robots.txt files.

You can contribute to it by speaking up and by protecting your work.

The takeaway

Bloggers did build a huge portion of what people now call the Internet’s knowledge. They explained the error messages, reviewed the gadgets, translated the legal jargon, and shared the recipes that nobody else bothered to write down. A lot of that work now feeds systems that answer questions without sending anyone back to the source.

You cannot fix that alone.

What you can do is stop relying on a single gatekeeper, build a direct line to your readers, write the kind of thing only you could write, and diversify how you earn. Be loud about attribution, careful about who crawls you, and patient about the long game.

The web was built by people who wrote things down and shared them. It is still worth doing. It just needs a different business model than the one we were promised.